We study equal and unequal-mass neutron star mergers by means of new numerical relativity simulations in which the general relativistic hydrodynamics solver employs an algorithm that guarantees mass conservation across the refinement levels of the computational mesh .
We consider eight binary configurations with total mass M = 2.7 M _ { \odot } , mass-ratios q = 1 and q = 1.16 , and four different equations of state ( EOSs ) , and one configuration with a stiff EOS , M = 2.5 M _ { \odot } and q = 1.5 , which is the largest mass ratio simulated in numerical relativity to date .
We focus on the post-merger dynamics and study the merger remnant , dynamical ejecta and the postmerger gravitational wave spectrum .
Although most of the merger remnant are a hypermassive neutron star collapsing to a black hole+disk system on dynamical timescales , stiff EOSs can eventually produce a stable massive neutron star .
During the merger process and on very short timescales , about \sim 10 ^ { -3 } -10 ^ { -2 } M _ { \odot } of material become unbound with kinetic energies \sim 10 ^ { 50 } \text { erg } .
Ejecta are mostly emitted around the orbital plane ; and favored by large mass ratios and softer EOS .
The postmerger wave spectrum is mainly characterized by the non-axisymmetric oscillations of the remnant neutron star .
The stiff EOS configuration consisting of a 1.5 M _ { \odot } and a 1.0 M _ { \odot } neutron star , simulated here for the first time , shows a rather peculiar dynamics .
During merger the companion star is very deformed ; about \sim 0.03 M _ { \odot } of rest-mass becomes unbound from the tidal tail due to the torque generated by the two-core inner structure .
The merger remnant is a stable neutron star surrounded by a massive accretion disk of rest-mass \sim 0.3 M _ { \odot } .
This and similar configurations might be particularly interesting for electromagnetic counterparts .
Comparing results obtained with and without the conservative mesh refinement algorithm , we find that post-merger simulations can be affected by systematic errors if mass conservation is not enforced in the mesh refinement strategy .
However , mass conservation also depends on grid details and on the artificial atmosphere setup ; the latter are particularly significant in the computation of the dynamical ejecta .